mango.image.registration.affine_transform

mango.image.registration.affine_transform(input, matrix, shift=None, offset=None, interptype=mango.image._mango_open_filters.InterpolationType.CATMULL_ROM_CUBIC_SPLINE, fill=None)

Applies an affine transformation to an image. This is the forward transform so that (conceptually):

idx  = scipy.array((i,j,k), dtype="int32")
tidx = matrix.dot((idx-offset).T) + offset + shift
out[tuple(tidx)] = input[tuple(idx)]

This differs from the scipy.ndimage.interpolation.affine_transform() function which does (conceptually, ignoring shift):

idx  = scipy.array((i,j,k), dtype="int32")
tidx = matrix.dot((idx-offset).T) + offset
out[tuple(idx)] = input[tuple(tidx)]

Parameters:

• input (mango.Dds) – Image to be transformed.
• matrix (numpy.ndarray) – A (3,3) shaped affine transformation matrix.
• shift (3-sequence) – The translation (number of voxels), can be float elements.
• offset (3-sequence) – The centre-point of the affine transformation (relative to input.origin). If None, the centre of the image is used as the centre of affine transformation. Elements can be float.
• interptype (mango.image.InterpolationType) – Interpolation type.
• fill (numeric) – The value used for elements outside the image-domain. If None uses the input.mtype.maskValue() or 0 if there is no input.mtype attribute.

Return type:

mango.Dds

Returns:

Affine-transformed mango.Dds image.